Use number-columns-repeated in ODS writer

The number-columns-repeated usage may reduce the size of the outputted XML file by merging repeated values together.

src/Spout/Writer/Common/Internal/AbstractWorkbook.php

@@ -135,7 +135,7 @@ abstract class AbstractWorkbook implements WorkbookInterface
      * If shouldCreateNewSheetsAutomatically option is set to true, it will handle pagination
      * with the creation of new worksheets if one worksheet has reached its maximum capicity.
      *
-     * @param array $dataRow Array containing data to be written.
+     * @param array $dataRow Array containing data to be written. Cannot be empty.
      *          Example $dataRow = ['data1', 1234, null, '', 'data5'];
      * @param \Box\Spout\Writer\Style\Style $style Style to be applied to the row.
      * @return void

src/Spout/Writer/ODS/Internal/Worksheet.php

@@ -7,7 +7,6 @@ use Box\Spout\Common\Exception\IOException;
 use Box\Spout\Common\Helper\StringHelper;
 use Box\Spout\Writer\Common\Helper\CellHelper;
 use Box\Spout\Writer\Common\Internal\WorksheetInterface;
-use Box\Spout\Writer\Common\Sheet;
 
 /**
  * Class Worksheet
@@ -126,7 +125,7 @@ class Worksheet implements WorksheetInterface
     /**
      * Adds data to the worksheet.
      *
-     * @param array $dataRow Array containing data to be written.
+     * @param array $dataRow Array containing data to be written. Cannot be empty.
      *          Example $dataRow = ['data1', 1234, null, '', 'data5'];
      * @param \Box\Spout\Writer\Style\Style $style Style to be applied to the row. NULL means use default style.
      * @return void
@@ -135,36 +134,26 @@ class Worksheet implements WorksheetInterface
      */
     public function addRow($dataRow, $style)
     {
-        $this->maxNumColumns = max($this->maxNumColumns, count($dataRow));
         $styleIndex = ($style->getId() + 1); // 1-based
+        $cellsCount = count($dataRow);
+        $this->maxNumColumns = max($this->maxNumColumns, $cellsCount);
 
         $data = '    <table:table-row table:style-name="ro1">' . PHP_EOL;
 
-        foreach($dataRow as $cellValue) {
+        $currentCellIndex = 0;
-            $data .= '        <table:table-cell table:style-name="ce' . $styleIndex . '"';
+        $nextCellIndex = 1;
 
-            if (CellHelper::isNonEmptyString($cellValue)) {
+        for ($i = 0; $i < $cellsCount; $i++) {
-                $data .= ' office:value-type="string" calcext:value-type="string">' . PHP_EOL;
+            $currentCellValue = $dataRow[$currentCellIndex];
 
-                $cellValueLines = explode("\n", $cellValue);
+            if (!array_key_exists($nextCellIndex, $dataRow) || $currentCellValue !== $dataRow[$nextCellIndex]) {
-                foreach ($cellValueLines as $cellValueLine) {
+                $numTimesValueRepeated = ($nextCellIndex - $currentCellIndex);
-                    $data .= '            <text:p>' . $this->stringsEscaper->escape($cellValueLine) . '</text:p>' . PHP_EOL;
+                $data .= $this->getCellContent($currentCellValue, $styleIndex, $numTimesValueRepeated);
-                }
 
-                $data .= '        </table:table-cell>' . PHP_EOL;
+                $currentCellIndex = $nextCellIndex;
-            } else if (CellHelper::isBoolean($cellValue)) {
-                $data .= ' office:value-type="boolean" calcext:value-type="boolean" office:value="' . $cellValue . '">' . PHP_EOL;
-                $data .= '            <text:p>' . $cellValue . '</text:p>' . PHP_EOL;
-                $data .= '        </table:table-cell>' . PHP_EOL;
-            } else if (CellHelper::isNumeric($cellValue)) {
-                $data .= ' office:value-type="float" calcext:value-type="float" office:value="' . $cellValue . '">' . PHP_EOL;
-                $data .= '            <text:p>' . $cellValue . '</text:p>' . PHP_EOL;
-                $data .= '        </table:table-cell>' . PHP_EOL;
-            } else if (empty($cellValue)) {
-                $data .= '/>' . PHP_EOL;
-            } else {
-                throw new InvalidArgumentException('Trying to add a value with an unsupported type: ' . gettype($cellValue));
             }
+
+            $nextCellIndex++;
         }
 
         $data .= '    </table:table-row>' . PHP_EOL;
@@ -178,6 +167,49 @@ class Worksheet implements WorksheetInterface
         $this->lastWrittenRowIndex++;
     }
 
+    /**
+     * Returns the cell XML content, given its value.
+     *
+     * @param mixed $cellValue The value to be written
+     * @param int $styleIndex Index of the used style
+     * @param int $numTimesValueRepeated Number of times the value is consecutively repeated
+     * @return string The cell XML content
+     * @throws \Box\Spout\Common\Exception\InvalidArgumentException If a cell value's type is not supported
+     */
+    protected function getCellContent($cellValue, $styleIndex, $numTimesValueRepeated)
+    {
+        $data = '        <table:table-cell table:style-name="ce' . $styleIndex . '"';
+
+        if ($numTimesValueRepeated !== 1) {
+            $data .= ' table:number-columns-repeated="' . $numTimesValueRepeated . '"';
+        }
+
+        if (CellHelper::isNonEmptyString($cellValue)) {
+            $data .= ' office:value-type="string" calcext:value-type="string">' . PHP_EOL;
+
+            $cellValueLines = explode("\n", $cellValue);
+            foreach ($cellValueLines as $cellValueLine) {
+                $data .= '            <text:p>' . $this->stringsEscaper->escape($cellValueLine) . '</text:p>' . PHP_EOL;
+            }
+
+            $data .= '        </table:table-cell>' . PHP_EOL;
+        } else if (CellHelper::isBoolean($cellValue)) {
+            $data .= ' office:value-type="boolean" calcext:value-type="boolean" office:value="' . $cellValue . '">' . PHP_EOL;
+            $data .= '            <text:p>' . $cellValue . '</text:p>' . PHP_EOL;
+            $data .= '        </table:table-cell>' . PHP_EOL;
+        } else if (CellHelper::isNumeric($cellValue)) {
+            $data .= ' office:value-type="float" calcext:value-type="float" office:value="' . $cellValue . '">' . PHP_EOL;
+            $data .= '            <text:p>' . $cellValue . '</text:p>' . PHP_EOL;
+            $data .= '        </table:table-cell>' . PHP_EOL;
+        } else if (empty($cellValue)) {
+            $data .= '/>' . PHP_EOL;
+        } else {
+            throw new InvalidArgumentException('Trying to add a value with an unsupported type: ' . gettype($cellValue));
+        }
+
+        return $data;
+    }
+
     /**
      * Closes the worksheet
      *

src/Spout/Writer/XLSX/Internal/Worksheet.php

@@ -118,7 +118,7 @@ EOD;
     /**
      * Adds data to the worksheet.
      *
-     * @param array $dataRow Array containing data to be written.
+     * @param array $dataRow Array containing data to be written. Cannot be empty.
      *          Example $dataRow = ['data1', 1234, null, '', 'data5'];
      * @param \Box\Spout\Writer\Style\Style $style Style to be applied to the row. NULL means use default style.
      * @return void

tests/Spout/Writer/ODS/WriterTest.php

@@ -195,6 +195,48 @@ class WriterTest extends \PHPUnit_Framework_TestCase
         $this->assertValueWasWrittenToSheet($fileName, 1, 10.2);
     }
 
+    /**
+     * @return array
+     */
+    public function dataProviderForTestAddRowShouldUseNumberColumnsRepeatedForRepeatedValues()
+    {
+        return [
+            [['ods--11', 'ods--11', 'ods--11'], 1, 3],
+            [['', ''], 1, 2],
+            [[true, true, true, true], 1, 4],
+            [[1.1, 1.1], 1, 2],
+            [['foo', 'bar'], 2, 0],
+        ];
+    }
+    /**
+     * @dataProvider dataProviderForTestAddRowShouldUseNumberColumnsRepeatedForRepeatedValues
+     *
+     * @param array $dataRow
+     * @param int $expectedNumTableCells
+     * @param int $expectedNumColumnsRepeated
+     * @return void
+     */
+    public function testAddRowShouldUseNumberColumnsRepeatedForRepeatedValues($dataRow, $expectedNumTableCells, $expectedNumColumnsRepeated)
+    {
+        $fileName = 'test_add_row_should_use_number_columns_repeated.ods';
+        $this->writeToODSFile([$dataRow], $fileName);
+
+        $sheetXmlNode = $this->getSheetXmlNode($fileName, 1);
+        $tableCellNodes = $sheetXmlNode->getElementsByTagName('table-cell');
+
+        $this->assertEquals($expectedNumTableCells, $tableCellNodes->length);
+
+        if ($expectedNumTableCells === 1) {
+            $tableCellNode = $tableCellNodes->item(0);
+            $numColumnsRepeated = intval($tableCellNode->getAttribute('table:number-columns-repeated'));
+            $this->assertEquals($expectedNumColumnsRepeated, $numColumnsRepeated);
+        } else {
+            foreach ($tableCellNodes as $tableCellNode) {
+                $this->assertFalse($tableCellNode->hasAttribute('table:number-columns-repeated'));
+            }
+        }
+    }
+
     /**
      * @return void
      */
@@ -423,12 +465,34 @@ class WriterTest extends \PHPUnit_Framework_TestCase
         $this->assertNotContains($valueAsXmlString, $sheetXmlAsString, $message);
     }
 
+    /**
+     * @param string $fileName
+     * @param int $sheetIndex
+     * @return \DOMNode
+     */
+    private function getSheetXmlNode($fileName, $sheetIndex)
+    {
+        $xmlReader = $this->moveReaderToCorrectTableNode($fileName, $sheetIndex);
+        return $xmlReader->expand();
+    }
+
     /**
      * @param string $fileName
      * @param int $sheetIndex
      * @return string
      */
     private function getSheetXmlNodeAsString($fileName, $sheetIndex)
+    {
+        $xmlReader = $this->moveReaderToCorrectTableNode($fileName, $sheetIndex);
+        return $xmlReader->readOuterXml();
+    }
+
+    /**
+     * @param string $fileName
+     * @param int $sheetIndex
+     * @return XMLReader
+     */
+    private function moveReaderToCorrectTableNode($fileName, $sheetIndex)
     {
         $resourcePath = $this->getGeneratedResourcePath($fileName);
         $pathToSheetFile = $resourcePath . '#content.xml';
@@ -441,6 +505,6 @@ class WriterTest extends \PHPUnit_Framework_TestCase
             $xmlReader->readUntilNodeFound('table:table');
         }
 
-        return $xmlReader->readOuterXml();
+        return $xmlReader;
     }
 }